linux-kernel - Re: [PATCH v3] riscv: Optimize crct10dif with zbc extension

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20250212195208.GA128826@sol.localdomain>
Date: Wed, 12 Feb 2025 11:52:08 -0800
From: Eric Biggers <ebiggers@...nel.org>
To: Zhihang Shao <zhihang.shao.iscas@...il.com>
Cc: linux-crypto@...r.kernel.org, linux-kernel@...r.kernel.org,
	linux-riscv@...ts.infradead.org, ardb@...nel.org
Subject: Re: [PATCH v3] riscv: Optimize crct10dif with zbc extension

On Wed, Feb 12, 2025 at 04:08:51PM +0800, Zhihang Shao wrote:
> On 2025/2/6 0:30, Eric Biggers wrote:
> 
> > Maybe use 'const __be64 *' and 'const __be32 *' for the pointer, and use
> > be64_to_cpu() and be32_to_cpu().  Then the __force cast won't be needed.
> Maybe this problem seems dumb, but I have no idea to adapt both riscv32 and
> riscv64. I have tried to use 'const __be64 *' and 'const __be32 *' pointer
> in patch v4, but I forgot to test them in riscv32, and it turns out the code
> failed to compile due to my mistake of defining 'const __be64 * p_ul' and of
> course it wouldn't be work for riscv32. Maybe I need some inspiration for
> this problem, or I still think it better to use 'unsigned long const *'
> since it works fine in both riscv64 and riscv32.
> 
> Looking forward to your reply and guidance.
> 
> Yours,
> 
> Zhihang
> 

I'm working on this patch and am going to send out a cleaned-up version.

- Eric